This project addresses a broad spectrum of interrelated biological problems that exploit structure based approaches to biochemical function prediction: Proposed improvements in protein structure and function prediction methodologies will lead to better approaches for drug lead selection and identification of of-target protein interactions responsible for drug side effects. Similarly, since DNA-protein interactions are crucial to a variety of biological processes including transcription, methods that better predict the identity and structure of DNA binding proteins in complex with DNA will be developed. Towards elucidation of the qualitative behavior of subcellular processes, schematic molecular simulations will be undertaken. A key idea is the observation that across evolutionarily distant proteins, there are conserved structural and functional features that can be used for ligand and protein ranking and refinement;remarkably, this conservation holds even at the level of protein side chain and ligand functional groups (e.g. hydroxyls). To realize these objectives, five Specific Aims are proposed: 1. TASSER will be improved and extended to treat models at atomic detail. 2. Exploiting structural conservation across multiple-domain holo-proteins, multidomain protein structure prediction will be improved. 3. Improved approaches for ligand ranking/identification of off-target proteins in proteomes will be developed. 4. Prediction of the DNA sequence dependence and quaternary structure of protein-bound DNA will be improved. 5. To explore the roles of hydrodynamic interactions (shown to be important for intracellular molecular diffusion) and crowding in intracellular dynamics, idealized Brownian Dynamics simulations of subcellular processes such as metabolic pathways, the mechanism of actin assembly, the mechanism of myosin motion along actin filaments, and aspects of protein dynamics associated with transcription will be undertaken. At every stage, careful and comprehensive benchmarking will be done;often, this will be followed by application of the developed methodology to proteomes. All tools and databases will be provided as web services;all software will be downloadable for use on local servers. Each Aim is part of a more comprehensive objective to increase our understanding of subcellular processes and biochemical function that will help accelerate drug discovery as well as provide general biochemical insights. PUBLIC HEALTH RELEVANCE: The development of more powerful computational methods for the prediction of protein structure, function and small molecule screening will improve the drug discovery process by providing better candidate drug leads as well as predicting possible off-target proteins that could give rise to drug side effects. The successful prediction of ligand cross-reactivity would significantly impact the development of safe and effective therapeutics. Finally, the simulations of subcellular processes will provide qualitative insights into how cells work and are the first tentative steps towards elucidating on a molecular level some of the intracellular changes giving rise to diseases such as cancer.